The confusion between "dominant" and "homozygous" is common and unfortunate. "Dominant" describes the mutation type, not the individual animal. Homozygous describes the genotype of an individual animal for a specific gene.
Genes generally come in pairs, one from each parent. A homozygous spider would have two spider mutant copies of the gene at the spider locus. Most if not all the spiders seen so far are only heterozygous for the spider gene, they only got the spider mutant copy from one parent and got a normal copy of that gene from the other parent.
A completely dominant mutation type would be defined as one where the homozygous genotype shows the same mutant appearance as the heterozygous genotype. Only one mutant copy of the gene is needed to completely dominate the normal copy and make the animal appear just as fully mutant as a homozygous mutant with two mutant copies.
Since the mutation type applies to the mutant gene and not the animal you see that pastel is still co-dominant regardless of if you are looking at a heterozygous pastel (the regular pastel phenotype) or a homozygous pastel (the super pastel phenotype). It is not correct to mix terminology and say that the super pastel is the dominant form - it's the homozygous form, but the pastel mutation is still the co-dominant type.
I find it easiest to predict the offspring by remembering the parent’s genotypes. Your spider X pastel cross is het spider X het pastel. That way you remember that the offspring each have a 50% chance of getting each mutation. Apparently these two mutations are on two different chromosomes, or at least not very close together if on the same, so the chances are fairly independent. Spider X Pastel should give you eggs each with about 25% chance of being bumblebee and the same for bumblebee X normal.
